Fix qemu-img create snapshot file failed when image format is raw
Corrected block stream command in human monitor (block-stream ->
block_stream)

changes from v1:
    Removed configuration to set QMP as default monitor
Signed-off-by: NXu Tian <xutian@redhat.com>

sg3_utils is requried by scsi_testsuite test,
so include it in unattended install files.
Signed-off-by: NXu Tian <xutian@redhat.com>
Acked-by: NXiaoqing Wei <xwei@redhat.com>
Signed-off-by: NQingtang Zhou <qzhou@redhat.com>

Currently we use autopart in all rhel guests, which bring
trouble when RHEL5 RHEL6 guests w/ huge(64+G) RAM. The
swap partition will be really huge and not enough space
for the root partition, thus installation fails.
(we met this problem on huge RAM host before).

With this patch, the RHEL5 RHEL6 guest will have only 2
physical partitions(/boot, PV), and the PV contains 2
 LVs(/, swap).

|     | /    |swap|
|/boot|----LVM----|
|     |    PV     |
|---------------- |
|guest hard drive |

Note:
1) we can use plain partitions here (w/o LVM) to gain the
   possible tiny performance improvement, but I prefer to
   choose the LVM scheme, a little bit complex partioning
   scheme but test more (also testing the lvm layer).
2) RHEL3, RHEL4, RHEL7 kickstart config didn't update here,
   because:
   The default kernel shipped w/ RHEL3, RHEL4 dont support
   such huge RAM, so they will not met this issue.
   The anaconda of RHEL7 looks fixed this problem, install
   w/ autopart will not create a huge swap partition.
Signed-off-by: NXiaoqing Wei <xwei@redhat.com>
Acked-by: NSuqin Huang <shuang@redhat.com>
Signed-off-by: NQingtang Zhou <qzhou@redhat.com>

qemu-guest-agent package exists in the cd image after RHEL6.3,
thus install it directly in guest installation stage.

For those RHEL6 guest before RHEL6.3, add a "ignoremissing"
option to let them ignore this new package.
Signed-off-by: NQingtang Zhou <qzhou@redhat.com>
Acked-by: NFeng Yang <fyang@redhat.com>

Signed-off-by: NQingtang Zhou <qzhou@redhat.com>

We just run some bash command in python. So directly run these
command in post script.
Signed-off-by: NFeng Yang <fyang@redhat.com>
Acked-by: NXiaoqing Wei <xwei@redhat.com>
Signed-off-by: NQingtang Zhou <qzhou@redhat.com>

Signed-off-by: NChris Evich <cevich@redhat.com>

Signed-off-by: NChris Evich <cevich@redhat.com>

Signed-off-by: NChris Evich <cevich@redhat.com>

Signed-off-by: Nwhuang <whuang@redhat.com>

Signed-off-by: Nwhuang <whuang@redhat.com>

Signed-off-by: Nwhuang <whuang@redhat.com>

Introduce virsh domif-setlink domif-getlink  testing:

    1. Positive testing
        1.1 Running domain and shutting off domain testing
        1.2 Setlink domain interface up and down then check
        1.3 Options "--config" testing
        1.4 Interface name and MAC address testing

    2. Negative testing
        2.1 Invalid options testing
        2.2 Shutting off domain with interface name
Signed-off-by: Nwhuang <whuang@redhat.com>

Signed-off-by: NYu Mingfei <yumingfei@cn.fujitsu.com>

Signed-off-by: Nyangdongsheng <yangds.fnst@cn.fujitsu.com>

(1).__schema_name__ = "nodedev"
(2).change get_key2syspath_dict and get_key2value_dict
    from @staticmethod to method of NodedevXML.
Signed-off-by: Nyangdongsheng <yangds.fnst@cn.fujitsu.com>

The default QEMU set, migrate, can usually be run with
a variety of QEMU shipped by current distros.

With the new migration protocols, people running the
default set will frequently have SKIPPED tests, which
is not good. So, restrict the default set to
protocols that are widely available. Once the new
formats are widespread (could take over a year for
that to happen), we can revert this change.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

When starting QEMU with an incoming migration protocol
it doesn't know about, a VMCreateError will be thrown.
We can do better in this case. Create a new exception
class that inherits from error.TestNAError, so the test
is skipped rather than errored or failed.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

Signed-off-by: NMichael R. Hines <mrhines@us.ibm.com>

This patch adds an extra directory level for RHEL guest,
so that we can add arch independent config in single place.
Signed-off-by: NQingtang Zhou <qzhou@redhat.com>

This `usb_storage` script would fail when getting
usb disk name in guest, since the guest doesn't
fully boot up.
Thus it needs some more time to wait for guest
initiate usb disk after boot.
Signed-off-by: NQingtang Zhou <qzhou@redhat.com>

If we run the usb_multi_disk.max_disk several times,
this test would fail because the images are formatted
and no usable free space.

This patch makes these images created force everytime,
to avoid this problem.
Signed-off-by: NQingtang Zhou <qzhou@redhat.com>

If there are more than one usb controllers on VM,
the usb port assignment code would assign one port
on every controller. This patch tries to fix this
problem.
Signed-off-by: NQingtang Zhou <qzhou@redhat.com>

The VMUSBControllerPortFullError exception doesn't provide
enough message to point out where has the problem.

This patch prints the usb device map when the exception is
raised, so that we can see which device caused the exception.
Signed-off-by: NQingtang Zhou <qzhou@redhat.com>

The virt-test framework checks both 'start_vm' and 'restart_vm'
parameters before starting a VM. So if we don't want to start
VM by framework, we should set both of these parameters to 'no'.
Signed-off-by: NQingtang Zhou <qzhou@redhat.com>

Signed-off-by: NJiří Župka <jzupka@redhat.com>

Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Sometimes we want to check qemu from a git tag, that allows
us to verify whether the code we are downloading actually
comes from the repository owner.

Introduce this support to our git helper. If for some reason
this verification fails, be merciless and fail the entire
build test.

CC: Michael S. Tirskin <mst@redhat.com>
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

Quite often the only difference between HMP and QMP monitor
output is missing or additional space. Add '' around outputs
to help finding those bugs.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

qemu_command has to be passed to qemu_has_option() function.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

This test previously only checked, if any CPU contains 'current=True'.
With this patch it goes through all cpus and checks, whether the
'current=True' is set only for the last CPU.

Additionally this verification method works with all smp setups.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

When start_vm was False, VM was not recreated, only params were
updated. Anyway when create() was called it didn't detect this
change and executed the previous command instead.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

chardev option should be wait instead of nowait and it requires
backend to be specified.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

init_pci_addr is non existing parameter.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>